Professional Documents
Culture Documents
3
(OH)
6
Cl
2
A. Zorko,
1,2
S. Nellutla,
3
J. van Tol,
3
L. C. Brunel,
3
F. Bert,
1
F. Duc,
4
J.-C. Trombe,
4
M. A. de Vries,
5
A. Harrison,
5
and P. Mendels
1
1
Laboratoire de Physique des Solides, UMR CNRS 8502, Universite Paris-Sud 11, 91405 Orsay, France
2
Jozef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia
3
National High Magnetic Field Laboratory, FSU, Tallahassee, Florida, USA
4
Centre dE
as the out-of-
plane and the in-plane components.
PRL 101, 026405 (2008)
P HYS I CAL RE VI E W L E T T E RS
week ending
11 JULY 2008
0031-9007,08,101(2),026405(4) 026405-1 2008 The American Physical Society
Using electron spin resonance (ESR), in this Letter we
provide evidence of a sizable magnetic anisotropy, which
we argue to be the DM interaction. In addition, through the
temperature dependence of the linewidth, we show that
spin correlations build up below 150 K and relate the
low-T behavior to intersite mixing defects.
Determined by the imaginary part of the dynamical
susceptibility (w), ESRspectra, I(w) (q - 0, w)
R
di(S
+
(i)S
(0))exp
|wi
,T, reect the T-evolution of the
spin correlation function (S
+
(i)S
ESR
1.0(2) emu,molCu at room temperature (RT),
which is in good agreement with the bulk susceptibility
l
1.1 emu,molCu. This proves that the Cu
2+
spins on
the kagome lattice are detected by ESR [18]. The
T-dependence of
ESR
is presented in Fig. 2(a). Its mono-
tonic increase with decreasing T closely resembles
l
down to 5 K and differs from the peaked kagome planes
susceptibility
k
[13,16]. This demonstrates that ESR de-
tects both the Cu
2+
spins on the kagome lattice and those
on the interlayer Zn sites, the latter resulting from the
Zn
2+
,Cu
2+
intersite disorder. Despite the two detected
Cu
2+
sites with likely different g-factors (ug & 0.2),
only a single ESR line is detected. This is expected [19]
because the exchange coupling 1
/
10 K between the two
Cu
2+
species [20] is larger than their difference in Zeeman
energy, ug
B
H & 1 K.
In order to t the spectra appropriately, one has to
consider both the g-factor anisotropy and the line width
broadening by the magnetic anisotropy. The former is
given by g(0) jg
2
|
cos
2
(0 0
g
) +g
2
J
sin
2
(0 0
g
)|
1,2
,
where 0 is the angle between H and the normal c to ka-
gome planes. 0
g
36
sin
2
0)
1,2
,
where H
z
and H
l
overshadows
k
, the Cu
2+
spins on the Zn sites domi-
nate the ESR response, which explains the crossover be-
is
allowed in ZnCu
3
(OH)
6
Cl
2
because the superexchange
mediating O
2
ions break the mirror symmetry of the
kagome planes. We therefore base our ESR analysis on
the dominance of the DM anisotropy. The D
|]
pattern
(Fig. 1) is obtained by choosing the direction of one of
the vectors and applying symmetry operators of the lattice
space group. We use the convention of spins being counted
clockwise in all triangles.
We can calculate the ESR line width [26]
H(0)
2r
_
k
l
2g(0)
B
1
j2D
2
z
+3D
2
+(2D
2
z
D
2
)cos
2
0|
3
16D
2
z
+78D
2
+(16D
2
z
26D
2
)cos
2
0
v
u
u
t
, (2)
determined by the DM pattern (Fig. 1), in the infinite-T
limit. The high-T exchange narrowing regime is reached at
RT since the ESR spectra do not change noticeably above
200 K. This is consistent with a T-independent NMR
spin-lattice relaxation above 150 K [16]. We tted the RT
ESR spectrum with the powder-averaged Lorentzian hav-
ing the above line width H(0). In Fig. 3, we plot on the
D
z
D
}
2(5) K. According to Eq. (2), their sign cannot be deter-
mined by ESR. We nd the magnitude of the extracted DM
interaction (D,1 - 0.08) in the range set by several other
Cu-based 2D frustrated systems. For instance, in the
orthogonal-dimer system SrCu
2
(BO
3
)
2
, it amounts to 4%
of the isotropic exchange [22,27], while the ratio of 16%
was reported for the triangular compound Cs
2
CuCl
4
[28].
At this point, it is important to address the issue of the
possible hidden symmetry of the DM interaction [29]. For
linear spin chains with staggered DM vectors, one can
effectively transform it to the AE term of the magnitude
D
2
,1 by applying a nonuniform spin rotation [30]. This
makes it of the same order as the initial AE and discards the
direct applicability of the KT formalism [30,31]. However,
for other lattices, one should distinguish between reducible
DM components, transforming to
2
order, and irreducible
components, remaining linear in [27]. The components
that can be eliminated in the rst order are those which sum
up to zero within any closed loop on the lattice [27]. For the
kagome lattice, the in-plane D
does not
signicantly inuence the ESR spectra (see Fig. 3). We
note that in SrCu
2
(BO
3
)
2
, the values of the DM interaction
obtained from the KT analysis [22], from inelastic neutron
scattering [27], and from NMR measurements [32] were
found to agree within 20%.
FIG. 3 (color online). The reduced
2
, reecting the quality of
the RT t in ZnCu
3
(OH)
6
Cl
2
. The dashed rectangle gives the
optimal parameters }D
z
} 15(1) K and }D
} 2(5) K.
PRL 101, 026405 (2008)
P HYS I CAL RE VI E W L E T T E RS
week ending
11 JULY 2008
026405-3
In Ref. [12], somewhat larger values, }D
z
} - 0.11 and
}D
M
3
2
,M
4
q
is for
k
B
T > g
B
H given by the second and the fourth mo-
ment, M
2
(jH
/
, S
+
|jS
, H
/
|),(S
+
S
) and M
4
(H H
Z
, jH
/
, S
+
|H H
Z
, jH
/
, S
|),(S
+
S
)
[24].
[27] Y. F. Cheng et al., Phys. Rev. B 75, 144422 (2007).
[28] R. Coldea et al., Phys. Rev. Lett. 88, 137203 (2002).
[29] T. A. Kaplan, Z. Phys. B 49, 313 (1983); L. Shekhtman,
O. Entin-Wohlman, and A. Aharony, Phys. Rev. Lett. 69,
836 (1992).
[30] J. Choukroun et al., Phys. Rev. Lett. 87, 127207 (2001).
[31] M. Oshikawa and I. Afeck, Phys. Rev. B 65, 134410
(2002).
[32] K. Kodama et al., J. Phys. Condens. Matter 17, L61
(2005).
[33] O. Cepas et al., arXiv:cond-mat/0806.0393 (to be pub-
lished).
[34] S. El Shawish (private communications).
[35] K. Gregor and O. I. Motrunich, Phys. Rev. B 77, 184423
(2008).
PRL 101, 026405 (2008)
P HYS I CAL RE VI E W L E T T E RS
week ending
11 JULY 2008
026405-4